修正建筑物内三维定位误差的运动感知方法研究

发布时间：2018-06-19 23:41

  本文选题：室内定位 + 累积误差　； 参考：《计算机应用研究》2017年09期

【摘要】：针对行人室内惯性定位存在航向发散和累积误差的问题,提出采用运动感知的方法实时识别行人的运动状态,利用行人状态约束(直线运动)和特殊运动状态(上下楼梯、乘坐电梯、90°或180°转弯)的特殊点匹配方法来抑制航向的发散,消除累积误差;提取加速度信号的时域特征,利用神经网络对行人运动状态进行识别,不同的运动形式(走、跑、上下楼梯)采取不同的步长模型,从而提高位置推算精度。为验证算法的有效性,在4楼和6楼之间进行多次行走实验,实验结果表明,通过运动感知可以明显抑制航向的发散,消除累积误差,三维空间内的定位误差约为总行程的1.6%。
[Abstract]:In order to solve the problem of heading divergence and accumulated error in indoor inertial positioning of pedestrians, a motion sensing method is proposed to identify the moving states of pedestrians in real time, using pedestrian state constraints (linear motion) and special motion states (up and down stairs). The special point matching method of 90 掳or 180 掳turn of elevator is used to restrain the divergence of course and eliminate the accumulated error, the time domain feature of acceleration signal is extracted, and the moving state of pedestrian is recognized by neural network, and the different motion forms (walking, running, running) are obtained. Different step-size models are used to improve the accuracy of position calculation. In order to verify the validity of the algorithm, several walking experiments were carried out between the fourth and sixth floors. The experimental results show that the divergence of course can be significantly suppressed by motion perception, and the accumulated error can be eliminated. The positioning error in three dimensional space is about 1.6 of the total stroke.
【作者单位】： 北京信息科技大学高动态导航技术北京市重点实验室;
【基金】：国家自然科学基金资助项目(61471046) 北京市教委市属高校创新能力提升计划资助项目(TJSHG201510772017) 高动态导航技术北京市重点实验室开放课题
【分类号】：TN96;TP183
，

本文编号：2041863